Blood-based markers for predicting malignancy in intraductal papillary mucinous neoplasms of the pancreas – what’s beyond the horizon?

Intraductal papillary mucinous neoplasm (IPMN) of the pancreas is a precursor lesion to pancreatic adenocarcinoma. It covers a broad histological spectrum, ranging from low grade dysplasia to invasive carcinoma. Being able to predict IPMN lesions with high-grade dysplasia or invasiveness that should undergo surgical resection is an important clinical task. However, identification of these high-risk lesions remains challenging as the decision is based mainly on radiological characteristics [1].

Cyst fluid and pancreatic juice represent promising sources for biomarker analysis [2]. However, the acquisition of these biofluids demands invasive procedures (e.g. EUS-FNA or ERCP) and comes with a complication risk [3–6]. Blood-based biomarkers may offer an alternative non-invasive strategy to improve the molecular diagnostics of IPMN. According to the recent European guidelines [7], there are no reliable blood-based biomarkers to predict high-grade dysplasia or invasiveness in pancreatic cystic neoplasms. Serum Carbohydrate Antigen 19-9 (CA 19-9) is to date the only clinically used marker in IPMN. However, the questions remain, how accurate is CA 19-9 in predicting individual malignancy risk in IPMN and are there any new blood-based biomarkers on the horizon? This editorial summarizes the recent progress in blood-based biomarkers for IPMN with a primary focus on biomarkers for predicting malignant pathology, as reported in the literature over the past 5 years.

Serum CA 19-9 is the most evaluated biomarker for predicting malignancy in IPMN. However, the diagnostic performance for predicting malignant IPMN, defined as high-grade dysplasia (formerly carcinoma in situ) and invasive carcinoma, is limited. Pooled analysis of 17 studies showed a sensitivity of 49%, a specificity of 89%, an area under the curve (AUC) of 0.78 and a diagnostic odds ratio of 7.29 [8].

Also serum CEA has limited diagnostic performance. Meta-analysis of 3 studies showed a sensitivity of 35%, a specificity of 95%, an AUC of 0.79 and a diagnostic odds ratio of 8.37 [8].

Serum CA 19-9 and CEA have been included in a nomogram together with cyst size, duct dilatation and mural nodule in order to evaluate risk of malignancy (high-grade dysplasia and associated invasive carcinoma) in branch-duct IPMN [9]. The diagnostic accuracy (AUC) for the whole nomogram was 0.737 among patients in Korea and Japan. External validation in Eastern and Western populations showed similar AUC [10].

Considering the moderate performance of serum CA 19-9 and CEA, additional investigational blood-based biomarkers have been evaluated for predicting malignancy in IPMN.

Neutrophil-to-lymphocyte ratio (NLR), a marker of systemic inflammatory response, has been investigated in several studies [11–15]. However, the optimal cut-off values remain controversial, with studies reporting NLR >2.074 (sensitivity 73%/specificity 58%) [11], NLR ≥2.2 (sensitivity 63%/specificity 86%) [12], NLR ≥2.5 (sensitivity 35%/specificity 87%) [13] and NLR >4 (sensitivity 34%/specificity 95%) [14] as significant predictive factors for high-grade dysplasia and invasive carcinoma [12,13], or only invasive carcinoma [11,14]. When evaluating NLR as a continuous variable it was found that NLR was significantly higher in invasive IPMN versus non-invasive IPMN [15]. However, no significant difference was seen between high-grade/invasive IPMN versus low/intermediate grade IPMN. The predictive performance of NLR can be improved when combined with CA 19-9, CEA and radiological findings [13].

MicroRNAs (miRNAs) and long non-coding RNAs (lncRNAs) are non-protein-coding transcripts that are involved in multiple cellular processes. A combination of five different miRNAs (miR-200a-3p, miR-1185-5p, miR-33a-5p, miR-574-3p and miR-663b) were found to be significantly correlated to risk of malignancy (high-grade dysplasia or invasive carcinoma) in IPMN [16]. The sensitivity was 81% with a specificity of 53%. In another study by the same research group, it was shown that 8 lncRNAs (ADARB2-AS1, ANRIL, GLIS3-AS1, LINC00472, MEG3, PANDA, PVT1 and UCA1) could predict malignant IPMN (high-grade dysplasia or invasive carcinoma) with a sensitivity of 77% and specificity of 76% [17]. When the lncRNA panel was combined with the miRNA data, radiomic data and standard clinical and radiologic features, the accuracy of predicting IPMN grade could be improved, yielding a sensitivity of 93% and specificity of 82% with an AUC of 0.92.

Changes in serum glycan profiles have been evaluated in order to predict the risk of invasive IPMN [18]. Nine glycans were found to be upregulated in invasive IPMN. The glycan m/z 3195 showed the highest diagnostic accuracy with a sensitivity of 92% and specificity of 67%.

Thrombospondin-2 (THBS2) is an anti-angiogenic glycoprotein that is involved in tumor progression. Circulating THBS2 levels were reported to be increased in high-grade/invasive IPMN lesions compared to low-risk IPMN lesions [19]. THBS2 alone provided an AUC of 0.65 for predicting IPMN grade. When THBS2 was included in a predictive model with CA 19-9, together with clinical and radiological parameters, the diagnostic accuracy increased to 0.82.

The insulin resistance markers leptin and branched chain amino acids were evaluated as blood markers for high-grade/invasive IPMN. Their diagnostic accuracy (AUC) was 0.61 [20]. When combined with CA 19-9 and main pancreatic duct diameter the predictive accuracy was increased to 0.81.

In summary, recent studies suggest that circulating biomarkers can be used in order to predict malignant IPMN. However, there is still insufficient evidence to support the use of these candidate biomarkers due to suboptimal predictive performance and lack of external validation. Current studies suggest that the diagnostic performance of investigational biomarkers can be improved when combined with tumor markers (e.g. serum CA 19-9 and CEA) and standard clinical and radiological features. In the future, with improved understanding of the molecular pathogenesis and natural history of IPMN and additional refinements of biomarker tools, it may become possible not only to predict malignancy, but also to detect the patients with a higher risk for developing a malignant transformation, and to operate these patients before the development of high-grade dysplasia/invasiveness.

Disclosure statement

No potential conflict of interest was reported by the author(s).